Preclinical perils and clinical catastrophe

When you get right down to it, everything from target discovery through model testing (animal or simulated human systems/organs) is “preclinical.” But practically speaking, I—and probably most of us—think of animal models when we think of preclinical testing. But increasingly, with ever-more-advanced in-silico models, organs-on-chips and whatever comes next, we are moving away from animals more and doing a better job of creating human-like systems that don’t use humans and don’t rely on “humanized” animals.

 

But we aren’t there yet. We aren’t clear of the use of animals, nor even the necessity to use animal models, however limited even the best ones might be. And even if we were further along, there are cultural, regulatory, bureaucratic, business and even scientific reasons why many would insist on not abandoning animal models altogether.

 

This in-between time that we are in does present problems, and we have seen those problems bloom most dramatically, like some horrific giant Venus flytrap, with chimeric antigen receptor T cell (CAR-T) therapy trials recently.

 

For example, in 2015, Juno Therapeutics launched a Phase 2 trial testing CAR-T therapy JCAR015—engineered with a specific protein to help immune cells identify and precisely kill tumor cells that displayed the CD19 antigen on their surface—for adult relapsed and refractory acute lymphoblastic leukemia (ALL). Although the therapy had shown great potential, by 2016 one of the 68 patients being treated with JCAR015 died from cerebral edema, followed by two more patients facing the same fate two months later. Though the trial was suspended, it resumed once Juno left out an accompanying chemotherapy drug it suspected was responsible for the adverse reactions. Then two more deaths from cerebral edema followed after that and, in early 2017, Juno pulled the plug on the trial and wrote off JCAR015 completely while continuing development on other CD19 therapies.

 

And it’s not as if JCAR015 is the only CAR-T cell therapy to have strong associations with serious patient morbidity and even significant mortality—ironically enough, death is even a specter (if you may forgive the pun) hanging over CAR-T cell therapy trials with highly successful outcomes. For example, Novartis’ candidate CLT019, which also targets CD19, demonstrated an outstanding 82-percent remission rate in a 2016 trial of children and young adults with ALL. However, in nearly half those patients the treatment caused severe cytokine release syndrome (CRS), a potentially lethal condition arising when a large amount of T cell-released cytokines set off inflammation and fever. Similarly, during a recent trial of its CD19-targeting investigational therapeutic KTE-C19 for non-Hodgkin lymphoma, Kite Pharma reported that serious neurological side effects occurred in one-third of its patients and CRS in one-fifth of them, not to mention two patient deaths.

 

And I bring all this up largely because of an email I received recently from Center for Responsible Science, a nonpartisan, nonprofit organization that works to improve lives by advancing regulatory science for modern and efficient drug and device development.

 

As the Center for Responsible Science noted, with at least 20 treatment-related deaths in clinical drug trials between May 2016 and April 2017, it has updated its July 2015 citizen petition, also adding a declaration from the father of a 24-year-old patient who was one of the Juno trial participants that died.

 

That amendment urges the U.S. Food and Drug Administration to amend 29 regulations to allow the preclinical test method most predictive of human response to be used during drug testing. Current regulations mandate the use of animal models, the Center for Responsible Science explains, even if there are human-relevant tests that could predict toxicity that animal tests may miss.

 

As the center puts it, “Despite their potential as a breakthrough cancer cure, there are serious safety concerns related to CAR-T cell therapies. Lack of relevant animal models for safety testing has been exemplified by numerous serious adverse events in studies using CAR-T engineered cells. It is essential that this promising cancer therapy be tested in human-relevant test methods to more specifically determine the safety risks before it is tested in humans.

 

“There are human-relevant test methods that can predict cytokine release syndrome and inflammation-related adverse events, including cerebral edema. CAR-T therapies could be tested in this platform in combination with any preconditioning drugs. However, due to current FDA regulations, the animal tests are required. It’s clear that traditional animal tests could not predict the deadly cerebral edema that killed five in the Juno ROCKET trial. Given what is known, [we] strongly encourage FDA to allow the test that most reliably predicts what will happen in humans, instead of insisting only on long-standing animal models.”

 

Here is where I, in my often maddening middle-of-the-road fashion, will both agree and disagree with the Center for Responsible Science and other organizations that co-sign their concerns in this area.

 

I’m not sure that FDA’s “old school” approach can be entirely blamed here. Would there have even been any reason to think that these particular human-relevant tests would be needed before the problems actually cropped up? The companies certainly weren’t expecting CRS and cerebral edema when they forged ahead. And given the cost of pharma and biotech R&D already, there is only so much we can expect in terms of testing. I don’t think we can ever be “exhaustive” and head all threats off at the pass.

 

Sure, now that we know these therapies pose the risk of things like CRS, I completely agree that the FDA needs to amend its rules to allow human-relevant tests in addition to or in lieu of animal tests for these kinds of therapies. But the implication that the wider reliance on animal model data is flawed may be a step too far—at least for now. I don’t think a blanket rule allowing human-relevant model data sounds right, given that many of those kinds of models are still relatively new on the scene.

 

Animal models are flawed and limited, certainly, but they also have a long track record and they do continue to improve.

 

I am among the first to admit that the FDA is slow to change (partly because it is not funded or staffed sufficiently to do all that we need it to do on even the “drug” part alone, much less the “food” part). It’s a bureaucracy, and that is pretty much par for the course in such institutions. The status quo is often much more enticing and easier to deal with than change. Change is needed, that I admit, and FDA almost certainly needs more pep in its step with regard to working human-relevant tests and models into the preclinical mix. There needs to be clear movement toward allowing such data as a boost to animal research or in some cases to one day replace animal research entirely.

 

But let’s also not forget the demands that researchers and patients (along with their families) both put on the FDA: Move faster with approvals; however, keep us safe from harm in clinical trials.

 

The FDA and clinical researchers can do better on the safety front. And they are. Already with CAR-T therapies, many companies are building in “kill switches” on the drugs so that they can monitor patients for the beginnings of things like CRS and turn off the therapy quickly if the signs start pointing toward harm. And putting better tests and models into play will help too.

 

But I don’t believe we can ever have “safe” clinical trials. The very purpose of such trials is to see how the therapies behave in humans, and that will always mean risk; that risk won’t be removed by human-relevant tests. Trials will often be the way we discover what the risks are so that they can be safer moving forward, as with CAR-T. It’s an uncomfortable truth but one with which we must live—and frankly, given that so many people in the Juno trial, for example, faced a chance of death in five years from ALL that was around 90 percent, I’m not sure how many would opt out even with the risk of CRS.

 

Let’s decrease the risks and increase the accuracy, but let’s never forget that trials are inherently risky, especially in areas like cancer where patients are desperate for cures.